Visual learning: diagrams you can click to the source

Not decoration — teaching tools

Novus Learn treats motion and interactivity as ways to teach, not as ornament. A diagram only ever shows relationships the source actually supports, and every clickable element opens that source's evidence rather than a guess.

The test we hold visuals to is simple: does this let you see something that the prose made you work to assemble? A picture that merely repeats the sentence above it has earned nothing. A picture that shows you the shape of a process, or which of six concepts is central, has done real work.

A shape for the idea

Different ideas want different pictures, so the planner recommends from the source's own structure rather than defaulting to one house style. Eight shapes are available:

• Network — related concepts and the labelled links between them.
• Timeline — dated events in the order the source gives them.
• Process — ordered steps that run start to finish.
• Cycle — steps that return to their beginning.
• Comparison — two things placed side by side on the source's own terms.
• Hierarchy — a strict tree, validated to make sure it never loops back on itself.
• Labeled system — the parts of a thing and the layers they sit in.
• Scale — magnitudes stated by the source, in units that genuinely compare.

The rules that keep a diagram honest

Each shape carries its own validation, and the checks are the interesting part. A hierarchy has to be an actual tree — a cycle in the parent links is rejected rather than drawn. A scale diagram may only use magnitudes the source states, in units from the same family, so nothing is silently converted and no invented reference object is added for a sense of scale. A labelled system is laid out abstractly rather than at invented coordinates, because the source described parts, not positions.

Click a node to peek at what it means and where it came from, in a small closable frame, so you can explore without losing your place. If a topic's structure doesn't support a shape, that shape isn't offered — an empty recommendation is better than a diagram that implies a relationship nobody claimed.

How a shape gets chosen

The planner reads the extracted structure rather than the prose. Dated entities in a sequence suggest a timeline. Stages that return to their first step suggest a cycle rather than a process — and getting that distinction right matters, because a cycle drawn as a line teaches the wrong thing about the subject.

Two entities with several contrasting claims each suggest a comparison. Parent-child relationships that form a clean tree suggest a hierarchy. Quantities stated with compatible units suggest a scale. When more than one shape fits, you get more than one recommendation and can pick; when none fits, you get none.

You can also build one yourself in the Visual Studio, choosing the type and arranging the nodes. The same validation applies to a diagram you made as to one the planner suggested, which occasionally means being told that the picture you wanted isn't one the source supports.

Watch it, don't just read it

Staged visuals can be played through with a scene player, which turns a process into something you step rather than scan. There's a presentation mode too, for putting a diagram on a projector without the rest of the interface coming along.

Everything respects your reduced-motion preference. With motion off, each scene renders complete and still, with the full text alternative intact — the content is never carried by the animation, so nothing is lost by turning it off. Auto-advance stays off in that mode as well.

That constraint shaped the visuals more than it sounds like it would. Because a scene has to be comprehensible as a single still frame, it can't rely on a reveal for its meaning — which turns out to be a good discipline generally. If a diagram only makes sense once you've watched it move, it usually means the still version was doing too little work in the first place.

When a visual isn't the answer

Some topics simply don't have a shape. An article that is mostly definitions produces a network of loosely-related nodes, which looks like a diagram without being one. Sources that describe change without dating it produce no timeline, because there is nothing to order it by.

There's also a ceiling on usable complexity. A diagram is capped at a few dozen nodes, and long before that limit a network stops being readable — at which point the right move is to study the sections separately rather than to look for a bigger canvas. When a topic resists visualisation, that is usually information about the topic.

And a diagram is a map of one source, not of a field. It shows what this article claimed, with this article's emphases and omissions. Read it that way and it's a genuinely fast way to see the shape of an argument; read it as a picture of the subject itself and it will quietly mislead you, which is precisely why every node stays one click from the sentence that put it there.